- Email: [email protected]
Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator
Journal of Orthopaedic Science xxx (xxxx) xxx
Contents lists available at ScienceDirect
Journal of Orthopaedic Science journal homepage: http://www.elsevier.com/locate/jos
Original Article
Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator Nathaniel T. Ondeck a, Michael C. Fu a, Ryan P. McLynn b, Patawut Bovonratwet a, Rohil Malpani c, Jonathan N. Grauer c, * a b c
Department of Orthopaedic Surgery, Hospital for Special Surgery, 535 E 70th Street, New York, NY, 10021, USA Department of Orthopaedic Surgery, University of Alabama at Birmingham School of Medicine, 1313 13th Street South, Birmingham, Al, 35205, USA Department of Orthopaedics and Rehabilitation, Yale School of Medicine, 800 Howard Avenue, New Haven, CT, 06510, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 July 2019 Accepted 25 September 2019 Available online xxx
Background: The last decade has seen increasing initiatives to improve health care delivery while decreasing financial expenditures, as particularly exemplified by the implementation of bundled payments for lower extremity arthroplasty, which hold the providers responsible for the both the quality and cost of these procedures. In this context, the utility of routine preoperative laboratory testing is unknown. The present study characterizes the associations, if any, between preoperative sodium, blood urea nitrogen (BUN), and creatinine values and the occurrence of general health adverse outcomes following total hip arthroplasty (THA). Methods: Patients undergoing primary THA were identified in the 2011e2015 National Surgical Quality Improvement Program. Cases with traumatic, oncologic, or infectious indications were excluded. Preoperative levels of sodium, BUN, and creatinine were tested for associations with perioperative adverse events and adverse hospital metrics using multivariate regressions that adjusted for patient baseline characteristics. Results: A total of 92,093 patients were included, of which 5.25% had an abnormal preoperative sodium level, 24.20% had an abnormal preoperative BUN level, and 11.95% had an abnormal preoperative creatinine level. Abnormal preoperative sodium levels (odds ratios: 1.23e1.50, p < 0.007) and creatinine levels (odds ratios: 1.27e1.55, p < 0.007) were associated with the occurrence of all studied adverse outcomes and abnormal preoperative BUN levels (odds ratios: 1.15e1.52, p < 0.007) were associated with the occurrence of all adverse outcomes except for hospital readmission. Conclusions: Abnormal preoperative laboratory testing is significantly associated with adverse outcomes following THA, supporting the added value of laboratory evaluation of patients before elective arthroplasty procedures. Study design: Clinical, Level III. © 2019 The Japanese Orthopaedic Association. Published by Elsevier B.V. All rights reserved.
1. Introduction With major surgical procedures such as lower extremity arthroplasty, value (defined as the quotient of quality over cost) is an increasingly scrutinized benchmark [1]. Perhaps this was best demonstrated when the Center for Medicare and Medicaid Services
* Corresponding author. Fax: þ203 785 7132. E-mail addresses: [email protected] (N.T. Ondeck), michael.fu.md@ gmail.com (M.C. Fu), [email protected] (R.P. McLynn), bovonratwetp@hss. edu (P. Bovonratwet), [email protected] (R. Malpani), jonathan.grauer@yale. edu (J.N. Grauer).
(CMS) in the United States launched the Bundled Payments for Care Improvement (BPCI) initiative in 2013 followed by the Comprehensive Care for Joint Replacement (CJR) program in 2015 [1,2]. These programs are pioneering several arrangements of episodebased reimbursement models in which health care systems are monetarily rewarded for clinical episode costs below a target price and penalized for clinical episode costs above a target price. Preliminary studies have noted decreased costs with the use of these alternative payment models while maintaining or improving the quality of care [3]. One method of addressing cost containment in such bundles is to create a standardized clinical pathway for each episode of care,
https://doi.org/10.1016/j.jos.2019.09.019 0949-2658/© 2019 The Japanese Orthopaedic Association. Published by Elsevier B.V. All rights reserved.
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
2
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
with the goal of reducing nonessential services [4]. With respect to total hip arthroplasty (THA), this can include adjusting preoperative routines, intraoperative procedures, and postoperative care. A balance must be struck between maintaining quality and safety while minimizing unnecessary aspects of care. Preoperative testing is an area that needs to be considered in this manner. For example, some hospitals have removed a routine preoperative urinalysis (unless the patient is symptomatic) and coagulation profile (unless a patient is currently being treated with anticoagulants or has a preexisting blood and/or liver disorder) [4]. An additional routine preoperative laboratory test is the basic metabolic panel, which includes sodium, blood urea nitrogen (BUN), and creatinine. While there are some studies that have conducted limited investigations of the utility of preoperative sodium [5e7], BUN [5e7], and creatinine levels [5e7], the generalizability of their conclusions are limited by the use of either local cohorts and/or a lack of a pure lower extremity joints population. To our knowledge, no study has provided a detailed investigation of associations of all three of these laboratory abnormalities with adverse outcomes following THA in a national cohort. It remains unclear if preoperative assessment of these three laboratory values is a valuable prognostic factor or is an artifact of the historic practice of medicine. In this context, the purpose of the present study is to identify the incidence of general health adverse outcomes across the spectrum of preoperative sodium, BUN and creatinine levels in a large, national dataset. Additionally, the present study aims to characterize specific associations, if any, between abnormally high and low preoperative sodium, BUN, and creatinine with perioperative adverse events and adverse hospital metrics following THA.
Preoperative laboratory values studied included sodium, BUN, and creatinine. These were characterized as both integer variables and also as categorical variables. The categories for all three lab values included abnormally low (sodium: <135 mEq/L, BUN: <7 mg/dL, creatinine: <0.5 mg/dL), normal (sodium: 135e145 mEq/L, BUN: 7e20 mg/dL, creatinine: 0.5e1.2 mg/dL), and abnormally high (sodium: >145 mEq/L, BUN: >20 mg/dL, creatinine > 1.2 mg/dL). These definitions were chosen as these have been previously used in the orthopaedic NSQIP literature [5].
2. Materials and methods
2.4. Statistical analysis
2.1. Patient sample
The incidence of all adverse outcome measures was calculated as a moving average involving the incidence of the preoperative value of interest as well as the incidence of preoperative values numerically two above and two below. Preoperative laboratory values on the high and low end of the spectrum (defined as an incidence of less than 0.1%), were combined. Chi-squared tests were used to compare the incidence of adverse outcomes for different preoperative laboratory value levels. Multivariate logistic regressions were employed to assess for the impact of abnormally low and abnormally high preoperative laboratory values on the studied adverse outcomes. All regressions controlled for all tabulated patient factors (age, gender, BMI, and ASA) and other studied preoperative laboratory values (sodium, BUN, and creatinine). Variance inflation factors (VIF) were calculated for all variables in the regression to alleviate concerns regarding collinearity (defined as a VIF > 10), as has been previously suggested in the statistical literature [14,15]. Missing preoperative laboratory data were handled with multiple imputation using chained equations, as multiple imputation is believed to enhance the robustness of evaluations in comparison with other methods, such as the commonly used complete case analysis [16,17]. All statistical tests were conducted using Stata version 13.1 (Stata Corp LLP, College Station, TX). A Bonferroni correction was employed to set the level of significance at p < 0.007. This study was granted an exemption from review by the authors’ institutional review board.
Patients with a Current Procedural Terminology code for primary THA (27130) were extracted from the National Surgical Quality Improvement Program (NSQIP) years 2011 through 2015. Patients undergoing revision procedures and/or who had traumatic, infectious, or oncologic indications were excluded. NSQIP is a quality improvement program that prospectively collects data on over 270 preoperative, intraoperative, and postoperative variables for patients undergoing a variety of surgical procedures from over 600 sites across the United States. Major advantages of this dataset include longitudinal tracking of patients for 30-days after their procedure, regardless of their admission status. Furthermore, the NSQIP dataset is chart-abstracted by trained surgical clinical reviewers, which is thought to offer significant advantages in data quality in comparison with administratively coded data [8,9]. NSQIP undergoes regular quality audits and the overall agreement rate of the reviews is approximately 2% for all assessed variables [10]. The strengths of this dataset have led to NSQIP being used extensively in arthroplasty research [7,11e13]. 2.2. Patient variables Patient age, sex, height, weight, and American Society of Anesthesiologists (ASA) Physical Status Classification were tabulated for all patients included in the study. Age was divided into commonly used subgroupings (18e39 years, 40e59 years, 60e79 years, 80 years). Height and weight were used to calculate body mass index (BMI), which was also divided into commonly used subcategories (<25 kg/m2, 25e29 kg/m2, 30e34 kg/m2, and 35 kg/m2). Patient ASA scores were grouped as Class 1/2 or as Class 3.
2.3. Patient outcomes General health adverse outcomes were divided into groupings of individual adverse events. These were severe adverse events (the occurrence of cardiac arrest, death, myocardial infarction, postoperative intubation, return to the operating room, sepsis, stroke, or venous thromboembolism), minor adverse events (the occurrence of acute kidney injury, anemia requiring transfusion, pneumonia, surgical site infection, urinary tract infection, or wound dehiscence), and any adverse event (the occurrence of either a severe or a minor adverse event). Adverse hospital metrics studied included the occurrence of an extended hospital stay (defined as length of stay greater than the 75th percentile, in this case, greater than 3 days), discharge to higher-level care (discharge to rehabilitation, a separate acute care, or a skilled/unskilled care facility that was not home), and hospital readmission. All adverse outcomes included in this study are tracked/reported in NSQIP for 30 days after the procedure. Adverse event and adverse hospital metrics similar to the ones described above have been used previously in the NSQIP orthopaedic literature [11e13].
3. Results In total, 92,093 patients were included in the study. The most common age group was 60e79 years of age (n ¼ 52,814, 57.35%),
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
there were slightly more females than males (n ¼ 50,707, 55.10%) the most common BMI category was 25e29 kg/m2 (n ¼ 31,022, 33.84%), and over half of patients were of ASA Class 1 or 2 (n ¼ 54,302, 59.03%). Demographic data is presented in Table 1. In terms of preoperative sodium levels, 4.50% (n ¼ 4,141) of patients had abnormally low values and 0.75% (n ¼ 690) of patients had abnormally high values. For preoperative BUN levels, 0.85% (n ¼ 780) of patients had abnormally low values, and 23.35% (n ¼ 21,502) of patients had abnormally high values. Finally, for preoperative creatinine values, 1.14% (n ¼ 1,047) had abnormally low values and 10.81% (n ¼ 9,959) of patients had abnormally high values. Laboratory data is also presented in Table 1. In total, 14.92% (n ¼ 13,736) of patients experienced any adverse event, 3.18% (n ¼ 2,925) of patients experienced a severe adverse event, and 12.90% (n ¼ 11,880) of patients experienced a minor adverse event. In terms of adverse hospital metrics, 16.70% (n ¼ 15,380) of patients had an extended hospital stay, 24.11% (n ¼ 22,205) of patients were discharged to a higher level of care, and 3.64% (n ¼ 3,352) of patients were readmitted within 30 days of their procedure. To evaluate the potential correlation of laboratory studies and adverse outcomes, any adverse event was selected as a representative aggregated adverse event measure and extended hospital stay was selected as a representative hospital metric measure. An evaluation of preoperative sodium revealed that the lowest incidence in the normal range for the occurrence of any adverse event (14.3%) and extended hospital stay (16.0%) occurred at 143 mEq/L
Table 1 Patient demographics, laboratory values, and adverse outcomes. Total
Age 18e39 years 40e59 years 60e79 years 80 years Sex Female Body mass index <25 kg/m2 25e29 kg/m2 30e34 kg/m2 35 kg/m2 American Society of Anesthesiologists Class Class 1/2 Class 3 Basic Metabolic Panel Sodium (Preoperative) Normal (135e145 mEq/L) Abnormal Low (<135 mEq/L) Abnormal High (>145 mEq/L) Missing BUN (Preoperative) Normal (7e20 mg/dL) Abnormal Low (<7 mg/dL) Abnormal High (>20 mg/dL) Missing Creatinine (Preoperative) Normal (0.5e1.2 mg/dL) Abnormal Low (<0.5 mg/dL) Abnormal High (>1.2 mg/dL) Missing Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to a Higher Level Care Hospital Readmission
Number
Percent
92,093
100.00%
2121 27,065 52,814 10,092
2.30% 29.39% 57.35% 10.96%
50,707
55.10%
18,404 31,022 23,521 18,726
20.08% 33.84% 25.66% 20.42%
54,302 37,691
59.03% 40.97%
80,878 4141 690 6384
87.82% 4.50% 0.75% 6.93%
59,509 780 21,502 10,302
64.62% 0.85% 23.35% 11.19%
75,711 1047 9959 5376 13,736 2925 11,880
82.21% 1.14% 10.81% 5.84% 14.92% 3.18% 12.90%
15,380 22,205 3352
16.70% 24.11% 3.64%
3
and 140 mEq/L, respectively. As preoperative sodium values become incrementally higher, there was not a clear trend of increasing or decreasing rates of the occurrence of any adverse event or of extended hospital stay. As preoperative sodium values became incrementally lower, there was a trend of increasing adverse outcomes. The peak incidence of any adverse event (32.4%) and extended hospital stay (31.2%) occurred at preoperative sodium levels 127 mEq/L and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). The incidences of any adverse event and extended hospital stay for the entire spectrum of preoperative sodium values can be graphically seen in Fig. 1. Similar trends were found for the other adverse outcome measures and the data is summarized in the multivariate analyses presented in Table 2. A similar evaluation of preoperative BUN revealed that the lowest incidence for the occurrence of any adverse event (13.5%) and extended hospital stay (13.9%) occurred at 16 mg/dL and 14 mg/dL, respectively. As BUN values become incrementally higher, there was a trend of increasing rates of any adverse event and extended hospital stay. The peak incidence of any adverse event (37.4%) and extended hospital stay (39.8%) occurred at preoperative BUN levels of 40 mg/dL and 43 mg/dL, respectively, and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). As BUN values become incrementally lower, there was also a trend of increasing rates of any adverse event and extended hospital stay. The peak incidence of any adverse event (22.1%) and extended hospital stay (24.5%) occurred at preoperative BUN levels of 3 mg/dL and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). The incidences of any adverse event and extended hospital stay for the entire spectrum of preoperative BUN values can be graphically seen in Fig. 2. Similar trends were found for the other adverse outcome measures and the data is summarized in the multivariate analyses presented in Table 3. A similar evaluation of preoperative creatinine revealed that the lowest incidence for the occurrence of any adverse event (13.6%) and extended hospital stay (15.3%) occurred at 1.0 mg/dL. As creatinine values become incrementally higher, there was an increase in the occurrence of any adverse event and extended hospital stay. The peak incidence of any adverse event (35.6%) and extended hospital stay (43.1%) occurred at preoperative creatinine levels of 2.2 mg/dL and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). As creatinine values became incrementally lower, there was an increase in the rates of any adverse event and extended hospital stay. The peak incidence of any adverse event (21.1%) and extended hospital stay (21.6%) occurred at preoperative creatinine levels of 0.3 mg/dL and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). The incidences of any adverse event and extended hospital stay for the entire spectrum of preoperative creatinine values can be graphically seen in Fig. 3. Similar trends were found for the other adverse outcome measures and the data is summarized in the multivariate analyses presented in Table 4. Multivariate logistic regressions revealed that low preoperative sodium was associated with the occurrence of all studied adverse outcomes (odds ratios: 1.23e1.50, p < 0.001), however elevated preoperative sodium was not associated with the occurrence of any adverse outcomes (Table 2). In terms of preoperative BUN, low values were associated with the occurrence of an extended hospital stay (OR: 1.52, p < 0.001) and a discharge to higher level of care (OR: 1.29, p ¼ 0.006). Elevated preoperative BUN values were associated with the occurrence of all adverse outcomes except for hospital readmission (OR: 1.15e1.33, p < 0.005) (Table 3).
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
4
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
Fig. 1. This graphic displays the incidence of the occurrence of any adverse event and extended hospital stay when stratified by integer values of preoperative sodium levels.
Table 2 Impact of preoperative sodium level on risk for adverse outcomes. Adverse Outcomes
Low Sodium (Preoperative)
High Sodium (Preoperative)
Odds Ratio
95% CI
P-value
Odds Ratio
95% CI
P-value
Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to Higher Level Care Hospital Readmission
1.48 1.41 1.50
1.37e1.60 1.22e1.65 1.38e1.62
<0.001 <0.001 <0.001
1.03 1.17 0.99
0.83e1.27 0.79e1.72 0.79e1.25
0.814 0.437 0.952
1.44 1.23 1.40
1.34e1.56 1.14e1.32 1.22e1.62
<0.001 <0.001 <0.001
1.27 1.26 1.21
1.05e1.53 1.06e1.50 0.85e1.72
0.012 0.008 0.298
Shading indicates statistical significance (p < 0.007 with Bonferroni correction). CI ¼ confidence interval.
Fig. 2. This graphic displays the incidence of the occurrence of any adverse event and extended hospital stay when stratified by integer values of preoperative BUN levels.
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
5
Table 3 Impact of preoperative BUN level on risk for adverse outcomes. Adverse Outcomes
Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to Higher Level Care Hospital Readmission
Low BUN (Preoperative)
High BUN (Preoperative)
Odds Ratio
95% CI
P-value
Odds Ratio
95% CI
P-value
1.22 1.13 1.17
1.01e1.46 0.77e1.68 0.96e1.42
0.037 0.526 0.116
1.22 1.15 1.22
1.16e1.27 1.04e1.26 1.16e1.29
<0.001 0.004 <0.001
1.52 1.29 1.32
1.28e1.81 1.07e1.55 0.95e1.83
<0.001 0.006 0.095
1.20 1.33 1.09
1.15e1.26 1.28e1.38 1.00e1.18
<0.001 <0.001 0.056
Shading indicates statistical significance (p < 0.007 with Bonferroni correction). CI ¼ confidence interval.
Fig. 3. This graphic displays the incidence of the occurrence of any adverse event and extended hospital stay when stratified by integer values of preoperative creatinine levels.
Table 4 Impact of preoperative creatinine level on risk for adverse outcomes. Adverse Outcomes
Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to Higher Level Care Hospital Readmission
Low Creatinine (Preoperative)
High Creatinine (Preoperative)
Odds Ratio
95% CI
P-value
Odds Ratio
95% CI
P-value
1.44 1.47 1.38
1.25e1.67 1.09e2.00 1.18e1.62
<0.001 0.013 <0.001
1.45 1.27 1.48
1.36e1.54 1.14e1.43 1.38e1.58
<0.001 <0.001 <0.001
1.55 1.31 1.43
1.34e1.79 1.14e1.50 1.08e1.91
<0.001 <0.001 0.014
1.44 1.35 1.40
1.36e1.53 1.29e1.43 1.26e1.55
<0.001 <0.001 <0.001
Shading indicates statistical significance (p < 0.007 with Bonferroni correction). CI ¼ confidence interval.
Finally, with respect to preoperative creatinine, low values were associated with the occurrence of any adverse event, minor adverse events, extended hospital stay, and discharge to higher level care (OR: 1.31e1.55, p < 0.001). Elevated preoperative creatinine values were associated with the occurrence of all adverse outcomes (OR: 1.27e1.48, p < 0.001) (Table 4). 4. Discussion As healthcare continues to adopt advance evidence based treatment algorithms, the incremental value of tests and
procedures is receiving increased scrutiny. This is especially important given the implementation of the voluntary BPCI and mandatory CJR initiatives for joint replacements, which demand healthcare facilities to consider both the quality and cost of treatment [1]. It is estimated that up to 18 billion dollars is spent each year on preoperative laboratory testing and diagnostic studies in the United States [18], and the use of guidelines to inform preoperative evaluation is thought to be key to economical pre-operative preparations [18,19]. Yet, to our knowledge, the clinical value of routine pre-operative laboratory abnormalities remain unclear in modern
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
6
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
elective arthroplasty patients. In this context, the purpose of the present study was to identify trends in the incidence of adverse outcomes across the spectrum of preoperative sodium, BUN, and creatinine levels in a large, national THA patient sample and to characterize correlations between these abnormal lab levels and adverse outcomes. Interestingly, excluding elevated preoperative sodium, the lowest incidence of all studied adverse outcomes occurred well within the traditionally defined normal ranges for sodium, BUN, and creatinine (Figs. 1e3). Additionally, when examining the upper and lower limits of normal ranges for all laboratory values, the highest incidence of any adverse event was within 10% of the local minimum value for all adverse outcomes. These findings support expectations that “normal” basic metabolic panel values of representative markers correlate with unremarkable perioperative courses. In the analyses of preoperative sodium levels, multivariate regressions controlling for patient factors and preoperative BUN and creatinine levels, revealed that abnormally low sodium levels were associated with the occurrence of all studied adverse outcomes (Table 2). Hyponatremia may be associated with increased adverse outcomes due to hypo-osmolality being correlated with impaired immune functioning, which can lead to higher propensities for infectious complications and their associated adverse hospital metrics [20]. Additionally, hyponatremia has been associated with encephalopathy [21], and peripheral nerve malfunction [22], which could respectively lead to attention deficits and gait unsteadiness. This has been documented to occur in patients with mild chronic hyponatremia and could also lead to a variety of adverse events and hospital metrics [23]. Given the high effect sizes of hyponatremia, patients with this finding may benefit from optimization of their sodium values before their planned procedure. Although many mild abnormalities may be due to common medications (thiazide diuretics, angiotensin-converting enzyme inhibitors, etc), it has been suggested that if there is not an apparent cause, investigation of the abnormality would be useful [24]. Interestingly, the same inclusive multivariate regressions revealed no significant associations between abnormally elevated sodium levels and postoperative adverse outcomes (Table 2). Although hypernatremia has also been previously associated with neurological dysfunction and myocardial impairments its hypothesized that the small number of patients in this sample with hypernatremia, in combination with the small range of elevated sodium levels may lead to hypernatremia not being a risk factor in this study [25]. In the analyses of preoperative BUN and creatinine levels, multivariate regressions revealed that both abnormally low and abnormally high preoperative levels were associated with adverse outcomes (Tables 3 and 4). Elevated levels of both of these tests is classically a marker of impaired renal function, however it can also be elevated in a variety of other conditions ranging from dehydration to gastrointestinal blood loss to congestive heart failure [26]. Reduced BUN and creatinine levels, although rare, can be caused by malnutrition, underlying severe liver disease, or syndrome of inappropriate antidiuretic hormone [27]. Regardless of the direction of abnormality, surgeons should consider further workup of high and low preoperative BUN and creatinine levels, as some of the causes are potentially reversible and steps to correct the irregularity may improve a patient's probability of an uncomplicated recovery. For example, studies have found that prophylactic dialysis in patients undergoing coronary bypass with creatinine values higher than 2.5 mg/dL led to decreased perioperative morbidity and mortality in comparison to those not receiving it [28].
Strengths of the present study include the use of the NSQIP dataset. As the patient information is abstracted by trained surgical clinical reviewers using chart review, it is thought to be of higher quality than administratively coded resources, such as the National Inpatient Sample [9]. Additionally, NSQIP provides patient followup for 30 days after surgery, regardless of admission status, allowing for the study of post-discharge adverse events. This is especially important given that nearly all proposed bundled payments require responsibility for care beyond the inpatient hospital stay [1]. Finally, the use of multiple imputation in the present study is novel and important in addressing known concerns for missing laboratory values in NSQIP. Previous literature has suggested that multiple imputation should be the standard for treating missing values [16,17]. Limitations of the present study include that it is impossible to assess actions taken by surgeons in response to abnormal preoperative laboratory values (for example replacing electrolytes in patients with hyponatremia, ordering dialysis for patients with elevated BUN or creatinine, and/or postponing/ canceling surgery) as this would not be captured by NSQIP. Additionally, there is no way to study the associations of other values that are often included in a basic metabolic panel (potassium, chloride, and bicarbonate) on adverse outcomes as they are not reported in the dataset. Despite these limitations, the present study offers the most convincing evidence to date of the utility of preoperative basic metabolic panel testing for the elective THA population. In conclusion, the current study demonstrates that traditionally defined normal ranges of preoperative sodium, BUN, and creatinine generally correlate with a lower incidence of perioperative general health adverse outcomes in THA patients. Further, by defining clear increased odds of adverse events and adverse hospital metrics in the setting of abnormal laboratory studies, even when controlling for patient factors, the current study supports the importance of evaluating these markers in the preoperative THA population. More research will be needed to assess the corrective impact of addressing abnormalities in these laboratories preoperatively. Ethical review committee statement This study has been given an exemption from the senior author's Institutional Review Board under federal regulation 45 CFR 46.101(b) (4). Declaration of Competing Interest The authors have no ownership, patents, or participations with entities whose products or category of products are mentioned in the current manuscript. No funding was received for this study. The authors of the manuscript do declare the following financial conflicts unrelated to the current manuscript: One of the authors (JNG) reports the following financial activities outside the submitted work: past consultancy with Medtronic (Minneapolis, MN, USA), Bioventus (Durham, NC, USA), Stryker (Mahwah, NJ, USA), and TIDI products (Neenah, Wisconsin, USA); and participation in a clinical trial as a sub-investigator for Pfizer (New York City, NY, USA), Spinal Kinetics (Sunnyvale, CA, USA), and Orthofix (Lewisville, TA, USA). All other authors certify that he or she has no commercial associations that might pose a conflict of interest in connection with the submitted article. References [1] Siddiqi A, White PB, Mistry JB, Gwam CU, Nace J, Mont MA, Delanois RE. Effect of bundled payments and health care reform as alternative payment models
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10] [11]
[12]
in total joint arthroplasty: a clinical review. J Arthroplast 2017 Aug;32(8): 2590e7. Froimson MI, Rana A, White Jr RE, Marshall A, Schutzer SF, Healy WL, Naas P, Daubert G, Iorio R, Parsley B. Bundled payments for care improvement initiative: the next evolution of payment formulations: AAHKS Bundled Payment Task Force. J Arthroplast 2013 Sep;28(Suppl. 8):157e65. Alfonso A, Hutzler L, Robb B, Beste C, Blom A, Bosco J. Similar cost savings of bundled payment initiatives applied to lower extremity total joint arthroplasty can Be achieved applying both models 2 and 3. HSS J 2017 Oct;13(3): 267e70. Iorio R, Clair AJ, Inneh IA, Slover JD, Bosco JA, Zuckerman JD. Early results of medicare's bundled payment initiative for a 90-day total joint arthroplasty episode of care. J Arthroplast 2016 Feb;31(2):343e50. Lakomkin N, Sathiyakumar V, Dodd AC, Jahangir AA, Whiting PS, Obremskey WT, Sethi MK. Pre-operative labs: wasted dollars or predictors of post-operative cardiac and septic events in orthopaedic trauma patients? Injury 2016 Jun;47(6):1217e21. Borges A, Torres J, Sao Simao R, Cabral AT, Pinto R. Impact of preoperative analytical values on post-operative mortality rate of intertrochanteric fractures. Acta Med Port 2014 Mar-Apr;27(2):218e22. Bernstein DN, Keswani A, Ring D. Perioperative risk adjustment for total shoulder arthroplasty: are simple clinically driven models sufficient? Clin Orthop Relat Res 2017 Dec;475(12):2867e74. Samuel AM, Lukasiewicz AM, Webb ML, Bohl DD, Basques BA, Davis KA, Grauer JN. ICD-9 diagnosis codes have poor sensitivity for identification of preexisting comorbidities in traumatic fracture patients: a study of the National Trauma Data Bank. J Trauma acute care surg 2015 Oct;79(4): 622e30. McLynn RP, Geddes BJ, Cui JJ, Ondeck NT, Bovonratwet P, Shultz BN, Grauer JN. Inaccuracies in ICD coding for obesity would be expected to bias administrative database spine studies toward overestimating the impact of obesity on perioperative adverse outcomes. Spine 2018 Aug 01;(7):526e32. Program ACoSNSQI. User guide for the 2015 ACS NSQIP participant use data file (PUF). Chicago, IL. 2016. Bovonratwet P, Ondeck NT, Tyagi V, Nelson SJ, Rubin LE, Grauer JN. Outpatient and inpatient unicompartmental knee arthroplasty procedures have similar short-term complication profiles. J Arthroplast 2017 May 18;32(10). Bohl DD, Samuel AM, Basques BA, Della Valle CJ, Levine BR, Grauer JN. How much do adverse event rates differ between primary and revision total joint arthroplasty? J Arthroplast 2016 Mar;31(3):596e602.
7
[13] Bohl DD, Fu MC, Golinvaux NS, Basques BA, Gruskay JA, Grauer JN. The "July effect" in primary total hip and knee arthroplasty: analysis of 21,434 cases from the ACS-NSQIP database. J Arthroplast 2014 Jul;29(7):1332e8. [14] Menard SW. Applied logistic regression analysis. Sage Publications; 1995. [15] Hair JF. Multivariate data analysis. Prentice Hall; 1998. [16] Parsons HM, Henderson WG, Ziegenfuss JY, Davern M, Al-Refaie WB. Missing data and interpretation of cancer surgery outcomes at the American College of surgeons national surgical quality improvement program. J Am Coll Surg 2011 Sep;213(3):379e91. [17] Ondeck NT, Fu MC, Skrip LA, McLynn RP, Su EP, Grauer JN. Treatments of missing values in large national data affect conclusions: the impact of multiple imputation on arthroplasty research. J Arthroplast 2018 Mar;33(3):661e7. [18] Richman DC. Ambulatory surgery: how much testing do we need? Anesthesiol Clin 2010 Jun;28(2):185e97. [19] Fischer SP. Cost-effective preoperative evaluation and testing. Chest 1999 May;115(Suppl. 5):96se100s. [20] Jensen AG, Wachmann CH, Poulsen KB, Espersen F, Scheibel J, Skinhoj P, Frimodt-Moller N. Risk factors for hospital-acquired Staphylococcus aureus bacteremia. Arch Intern Med 1999 Jul 12;159(13):1437e44. [21] Hoorn EJ, Zietse R. Hyponatremia and mortality: moving beyond associations. Am J Kidney Dis 2013 Jul;62(1):139e49. [22] Aranyi Z, Kovacs T, Szirmai I, Vastagh I. Reversible nerve conduction slowing in hyponatremia. J Neurol 2004 Dec;251(12):1532e3. [23] Renneboog B, Musch W, Vandemergel X, Manto MU, Decaux G. Mild chronic hyponatremia is associated with falls, unsteadiness, and attention deficits. Am J Med 2006 Jan;119(1):71 e1e8. [24] Klinck J, McNeill L, Di Angelantonio E, Menon DK. Predictors and outcome impact of perioperative serum sodium changes in a high-risk population. Br J Anaesth 2015 Apr;114(4):615e22. [25] Adrogue HJ, Madias NE. Hypernatremia. N Engl J Med 2000 May 18;342(20): 1493e9. [26] Jafari SM, Huang R, Joshi A, Parvizi J, Hozack WJ. Renal impairment following total joint arthroplasty: who is at risk? J Arthroplast 2010 Sep;25(Suppl. 6): 49e53. e1-2. [27] Longo D, Fauci A, Kasper D, Hauser S, Jameson J, Loscalzo J. Harrison's principles of internal medicine. 18th ed. McGraw-Hill Education; 2011. [28] Durmaz I, Yagdi T, Calkavur T, Mahmudov R, Apaydin AZ, Posacioglu H, Atay Y, Engin C. Prophylactic dialysis in patients with renal dysfunction undergoing on-pump coronary artery bypass surgery. Ann Thorac Surg 2003 Mar;75(3): 859e64.
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
Contents lists available at ScienceDirect
Journal of Orthopaedic Science journal homepage: http://www.elsevier.com/locate/jos
Original Article
Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator Nathaniel T. Ondeck a, Michael C. Fu a, Ryan P. McLynn b, Patawut Bovonratwet a, Rohil Malpani c, Jonathan N. Grauer c, * a b c
Department of Orthopaedic Surgery, Hospital for Special Surgery, 535 E 70th Street, New York, NY, 10021, USA Department of Orthopaedic Surgery, University of Alabama at Birmingham School of Medicine, 1313 13th Street South, Birmingham, Al, 35205, USA Department of Orthopaedics and Rehabilitation, Yale School of Medicine, 800 Howard Avenue, New Haven, CT, 06510, USA
a r t i c l e i n f o
a b s t r a c t
Article history: Received 1 July 2019 Accepted 25 September 2019 Available online xxx
Background: The last decade has seen increasing initiatives to improve health care delivery while decreasing financial expenditures, as particularly exemplified by the implementation of bundled payments for lower extremity arthroplasty, which hold the providers responsible for the both the quality and cost of these procedures. In this context, the utility of routine preoperative laboratory testing is unknown. The present study characterizes the associations, if any, between preoperative sodium, blood urea nitrogen (BUN), and creatinine values and the occurrence of general health adverse outcomes following total hip arthroplasty (THA). Methods: Patients undergoing primary THA were identified in the 2011e2015 National Surgical Quality Improvement Program. Cases with traumatic, oncologic, or infectious indications were excluded. Preoperative levels of sodium, BUN, and creatinine were tested for associations with perioperative adverse events and adverse hospital metrics using multivariate regressions that adjusted for patient baseline characteristics. Results: A total of 92,093 patients were included, of which 5.25% had an abnormal preoperative sodium level, 24.20% had an abnormal preoperative BUN level, and 11.95% had an abnormal preoperative creatinine level. Abnormal preoperative sodium levels (odds ratios: 1.23e1.50, p < 0.007) and creatinine levels (odds ratios: 1.27e1.55, p < 0.007) were associated with the occurrence of all studied adverse outcomes and abnormal preoperative BUN levels (odds ratios: 1.15e1.52, p < 0.007) were associated with the occurrence of all adverse outcomes except for hospital readmission. Conclusions: Abnormal preoperative laboratory testing is significantly associated with adverse outcomes following THA, supporting the added value of laboratory evaluation of patients before elective arthroplasty procedures. Study design: Clinical, Level III. © 2019 The Japanese Orthopaedic Association. Published by Elsevier B.V. All rights reserved.
1. Introduction With major surgical procedures such as lower extremity arthroplasty, value (defined as the quotient of quality over cost) is an increasingly scrutinized benchmark [1]. Perhaps this was best demonstrated when the Center for Medicare and Medicaid Services
* Corresponding author. Fax: þ203 785 7132. E-mail addresses: [email protected] (N.T. Ondeck), michael.fu.md@ gmail.com (M.C. Fu), [email protected] (R.P. McLynn), bovonratwetp@hss. edu (P. Bovonratwet), [email protected] (R. Malpani), jonathan.grauer@yale. edu (J.N. Grauer).
(CMS) in the United States launched the Bundled Payments for Care Improvement (BPCI) initiative in 2013 followed by the Comprehensive Care for Joint Replacement (CJR) program in 2015 [1,2]. These programs are pioneering several arrangements of episodebased reimbursement models in which health care systems are monetarily rewarded for clinical episode costs below a target price and penalized for clinical episode costs above a target price. Preliminary studies have noted decreased costs with the use of these alternative payment models while maintaining or improving the quality of care [3]. One method of addressing cost containment in such bundles is to create a standardized clinical pathway for each episode of care,
https://doi.org/10.1016/j.jos.2019.09.019 0949-2658/© 2019 The Japanese Orthopaedic Association. Published by Elsevier B.V. All rights reserved.
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
2
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
with the goal of reducing nonessential services [4]. With respect to total hip arthroplasty (THA), this can include adjusting preoperative routines, intraoperative procedures, and postoperative care. A balance must be struck between maintaining quality and safety while minimizing unnecessary aspects of care. Preoperative testing is an area that needs to be considered in this manner. For example, some hospitals have removed a routine preoperative urinalysis (unless the patient is symptomatic) and coagulation profile (unless a patient is currently being treated with anticoagulants or has a preexisting blood and/or liver disorder) [4]. An additional routine preoperative laboratory test is the basic metabolic panel, which includes sodium, blood urea nitrogen (BUN), and creatinine. While there are some studies that have conducted limited investigations of the utility of preoperative sodium [5e7], BUN [5e7], and creatinine levels [5e7], the generalizability of their conclusions are limited by the use of either local cohorts and/or a lack of a pure lower extremity joints population. To our knowledge, no study has provided a detailed investigation of associations of all three of these laboratory abnormalities with adverse outcomes following THA in a national cohort. It remains unclear if preoperative assessment of these three laboratory values is a valuable prognostic factor or is an artifact of the historic practice of medicine. In this context, the purpose of the present study is to identify the incidence of general health adverse outcomes across the spectrum of preoperative sodium, BUN and creatinine levels in a large, national dataset. Additionally, the present study aims to characterize specific associations, if any, between abnormally high and low preoperative sodium, BUN, and creatinine with perioperative adverse events and adverse hospital metrics following THA.
Preoperative laboratory values studied included sodium, BUN, and creatinine. These were characterized as both integer variables and also as categorical variables. The categories for all three lab values included abnormally low (sodium: <135 mEq/L, BUN: <7 mg/dL, creatinine: <0.5 mg/dL), normal (sodium: 135e145 mEq/L, BUN: 7e20 mg/dL, creatinine: 0.5e1.2 mg/dL), and abnormally high (sodium: >145 mEq/L, BUN: >20 mg/dL, creatinine > 1.2 mg/dL). These definitions were chosen as these have been previously used in the orthopaedic NSQIP literature [5].
2. Materials and methods
2.4. Statistical analysis
2.1. Patient sample
The incidence of all adverse outcome measures was calculated as a moving average involving the incidence of the preoperative value of interest as well as the incidence of preoperative values numerically two above and two below. Preoperative laboratory values on the high and low end of the spectrum (defined as an incidence of less than 0.1%), were combined. Chi-squared tests were used to compare the incidence of adverse outcomes for different preoperative laboratory value levels. Multivariate logistic regressions were employed to assess for the impact of abnormally low and abnormally high preoperative laboratory values on the studied adverse outcomes. All regressions controlled for all tabulated patient factors (age, gender, BMI, and ASA) and other studied preoperative laboratory values (sodium, BUN, and creatinine). Variance inflation factors (VIF) were calculated for all variables in the regression to alleviate concerns regarding collinearity (defined as a VIF > 10), as has been previously suggested in the statistical literature [14,15]. Missing preoperative laboratory data were handled with multiple imputation using chained equations, as multiple imputation is believed to enhance the robustness of evaluations in comparison with other methods, such as the commonly used complete case analysis [16,17]. All statistical tests were conducted using Stata version 13.1 (Stata Corp LLP, College Station, TX). A Bonferroni correction was employed to set the level of significance at p < 0.007. This study was granted an exemption from review by the authors’ institutional review board.
Patients with a Current Procedural Terminology code for primary THA (27130) were extracted from the National Surgical Quality Improvement Program (NSQIP) years 2011 through 2015. Patients undergoing revision procedures and/or who had traumatic, infectious, or oncologic indications were excluded. NSQIP is a quality improvement program that prospectively collects data on over 270 preoperative, intraoperative, and postoperative variables for patients undergoing a variety of surgical procedures from over 600 sites across the United States. Major advantages of this dataset include longitudinal tracking of patients for 30-days after their procedure, regardless of their admission status. Furthermore, the NSQIP dataset is chart-abstracted by trained surgical clinical reviewers, which is thought to offer significant advantages in data quality in comparison with administratively coded data [8,9]. NSQIP undergoes regular quality audits and the overall agreement rate of the reviews is approximately 2% for all assessed variables [10]. The strengths of this dataset have led to NSQIP being used extensively in arthroplasty research [7,11e13]. 2.2. Patient variables Patient age, sex, height, weight, and American Society of Anesthesiologists (ASA) Physical Status Classification were tabulated for all patients included in the study. Age was divided into commonly used subgroupings (18e39 years, 40e59 years, 60e79 years, 80 years). Height and weight were used to calculate body mass index (BMI), which was also divided into commonly used subcategories (<25 kg/m2, 25e29 kg/m2, 30e34 kg/m2, and 35 kg/m2). Patient ASA scores were grouped as Class 1/2 or as Class 3.
2.3. Patient outcomes General health adverse outcomes were divided into groupings of individual adverse events. These were severe adverse events (the occurrence of cardiac arrest, death, myocardial infarction, postoperative intubation, return to the operating room, sepsis, stroke, or venous thromboembolism), minor adverse events (the occurrence of acute kidney injury, anemia requiring transfusion, pneumonia, surgical site infection, urinary tract infection, or wound dehiscence), and any adverse event (the occurrence of either a severe or a minor adverse event). Adverse hospital metrics studied included the occurrence of an extended hospital stay (defined as length of stay greater than the 75th percentile, in this case, greater than 3 days), discharge to higher-level care (discharge to rehabilitation, a separate acute care, or a skilled/unskilled care facility that was not home), and hospital readmission. All adverse outcomes included in this study are tracked/reported in NSQIP for 30 days after the procedure. Adverse event and adverse hospital metrics similar to the ones described above have been used previously in the NSQIP orthopaedic literature [11e13].
3. Results In total, 92,093 patients were included in the study. The most common age group was 60e79 years of age (n ¼ 52,814, 57.35%),
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
there were slightly more females than males (n ¼ 50,707, 55.10%) the most common BMI category was 25e29 kg/m2 (n ¼ 31,022, 33.84%), and over half of patients were of ASA Class 1 or 2 (n ¼ 54,302, 59.03%). Demographic data is presented in Table 1. In terms of preoperative sodium levels, 4.50% (n ¼ 4,141) of patients had abnormally low values and 0.75% (n ¼ 690) of patients had abnormally high values. For preoperative BUN levels, 0.85% (n ¼ 780) of patients had abnormally low values, and 23.35% (n ¼ 21,502) of patients had abnormally high values. Finally, for preoperative creatinine values, 1.14% (n ¼ 1,047) had abnormally low values and 10.81% (n ¼ 9,959) of patients had abnormally high values. Laboratory data is also presented in Table 1. In total, 14.92% (n ¼ 13,736) of patients experienced any adverse event, 3.18% (n ¼ 2,925) of patients experienced a severe adverse event, and 12.90% (n ¼ 11,880) of patients experienced a minor adverse event. In terms of adverse hospital metrics, 16.70% (n ¼ 15,380) of patients had an extended hospital stay, 24.11% (n ¼ 22,205) of patients were discharged to a higher level of care, and 3.64% (n ¼ 3,352) of patients were readmitted within 30 days of their procedure. To evaluate the potential correlation of laboratory studies and adverse outcomes, any adverse event was selected as a representative aggregated adverse event measure and extended hospital stay was selected as a representative hospital metric measure. An evaluation of preoperative sodium revealed that the lowest incidence in the normal range for the occurrence of any adverse event (14.3%) and extended hospital stay (16.0%) occurred at 143 mEq/L
Table 1 Patient demographics, laboratory values, and adverse outcomes. Total
Age 18e39 years 40e59 years 60e79 years 80 years Sex Female Body mass index <25 kg/m2 25e29 kg/m2 30e34 kg/m2 35 kg/m2 American Society of Anesthesiologists Class Class 1/2 Class 3 Basic Metabolic Panel Sodium (Preoperative) Normal (135e145 mEq/L) Abnormal Low (<135 mEq/L) Abnormal High (>145 mEq/L) Missing BUN (Preoperative) Normal (7e20 mg/dL) Abnormal Low (<7 mg/dL) Abnormal High (>20 mg/dL) Missing Creatinine (Preoperative) Normal (0.5e1.2 mg/dL) Abnormal Low (<0.5 mg/dL) Abnormal High (>1.2 mg/dL) Missing Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to a Higher Level Care Hospital Readmission
Number
Percent
92,093
100.00%
2121 27,065 52,814 10,092
2.30% 29.39% 57.35% 10.96%
50,707
55.10%
18,404 31,022 23,521 18,726
20.08% 33.84% 25.66% 20.42%
54,302 37,691
59.03% 40.97%
80,878 4141 690 6384
87.82% 4.50% 0.75% 6.93%
59,509 780 21,502 10,302
64.62% 0.85% 23.35% 11.19%
75,711 1047 9959 5376 13,736 2925 11,880
82.21% 1.14% 10.81% 5.84% 14.92% 3.18% 12.90%
15,380 22,205 3352
16.70% 24.11% 3.64%
3
and 140 mEq/L, respectively. As preoperative sodium values become incrementally higher, there was not a clear trend of increasing or decreasing rates of the occurrence of any adverse event or of extended hospital stay. As preoperative sodium values became incrementally lower, there was a trend of increasing adverse outcomes. The peak incidence of any adverse event (32.4%) and extended hospital stay (31.2%) occurred at preoperative sodium levels 127 mEq/L and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). The incidences of any adverse event and extended hospital stay for the entire spectrum of preoperative sodium values can be graphically seen in Fig. 1. Similar trends were found for the other adverse outcome measures and the data is summarized in the multivariate analyses presented in Table 2. A similar evaluation of preoperative BUN revealed that the lowest incidence for the occurrence of any adverse event (13.5%) and extended hospital stay (13.9%) occurred at 16 mg/dL and 14 mg/dL, respectively. As BUN values become incrementally higher, there was a trend of increasing rates of any adverse event and extended hospital stay. The peak incidence of any adverse event (37.4%) and extended hospital stay (39.8%) occurred at preoperative BUN levels of 40 mg/dL and 43 mg/dL, respectively, and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). As BUN values become incrementally lower, there was also a trend of increasing rates of any adverse event and extended hospital stay. The peak incidence of any adverse event (22.1%) and extended hospital stay (24.5%) occurred at preoperative BUN levels of 3 mg/dL and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). The incidences of any adverse event and extended hospital stay for the entire spectrum of preoperative BUN values can be graphically seen in Fig. 2. Similar trends were found for the other adverse outcome measures and the data is summarized in the multivariate analyses presented in Table 3. A similar evaluation of preoperative creatinine revealed that the lowest incidence for the occurrence of any adverse event (13.6%) and extended hospital stay (15.3%) occurred at 1.0 mg/dL. As creatinine values become incrementally higher, there was an increase in the occurrence of any adverse event and extended hospital stay. The peak incidence of any adverse event (35.6%) and extended hospital stay (43.1%) occurred at preoperative creatinine levels of 2.2 mg/dL and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). As creatinine values became incrementally lower, there was an increase in the rates of any adverse event and extended hospital stay. The peak incidence of any adverse event (21.1%) and extended hospital stay (21.6%) occurred at preoperative creatinine levels of 0.3 mg/dL and were statistically larger than the described above lowest incidence in the normal range (p < 0.001). The incidences of any adverse event and extended hospital stay for the entire spectrum of preoperative creatinine values can be graphically seen in Fig. 3. Similar trends were found for the other adverse outcome measures and the data is summarized in the multivariate analyses presented in Table 4. Multivariate logistic regressions revealed that low preoperative sodium was associated with the occurrence of all studied adverse outcomes (odds ratios: 1.23e1.50, p < 0.001), however elevated preoperative sodium was not associated with the occurrence of any adverse outcomes (Table 2). In terms of preoperative BUN, low values were associated with the occurrence of an extended hospital stay (OR: 1.52, p < 0.001) and a discharge to higher level of care (OR: 1.29, p ¼ 0.006). Elevated preoperative BUN values were associated with the occurrence of all adverse outcomes except for hospital readmission (OR: 1.15e1.33, p < 0.005) (Table 3).
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
4
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
Fig. 1. This graphic displays the incidence of the occurrence of any adverse event and extended hospital stay when stratified by integer values of preoperative sodium levels.
Table 2 Impact of preoperative sodium level on risk for adverse outcomes. Adverse Outcomes
Low Sodium (Preoperative)
High Sodium (Preoperative)
Odds Ratio
95% CI
P-value
Odds Ratio
95% CI
P-value
Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to Higher Level Care Hospital Readmission
1.48 1.41 1.50
1.37e1.60 1.22e1.65 1.38e1.62
<0.001 <0.001 <0.001
1.03 1.17 0.99
0.83e1.27 0.79e1.72 0.79e1.25
0.814 0.437 0.952
1.44 1.23 1.40
1.34e1.56 1.14e1.32 1.22e1.62
<0.001 <0.001 <0.001
1.27 1.26 1.21
1.05e1.53 1.06e1.50 0.85e1.72
0.012 0.008 0.298
Shading indicates statistical significance (p < 0.007 with Bonferroni correction). CI ¼ confidence interval.
Fig. 2. This graphic displays the incidence of the occurrence of any adverse event and extended hospital stay when stratified by integer values of preoperative BUN levels.
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
5
Table 3 Impact of preoperative BUN level on risk for adverse outcomes. Adverse Outcomes
Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to Higher Level Care Hospital Readmission
Low BUN (Preoperative)
High BUN (Preoperative)
Odds Ratio
95% CI
P-value
Odds Ratio
95% CI
P-value
1.22 1.13 1.17
1.01e1.46 0.77e1.68 0.96e1.42
0.037 0.526 0.116
1.22 1.15 1.22
1.16e1.27 1.04e1.26 1.16e1.29
<0.001 0.004 <0.001
1.52 1.29 1.32
1.28e1.81 1.07e1.55 0.95e1.83
<0.001 0.006 0.095
1.20 1.33 1.09
1.15e1.26 1.28e1.38 1.00e1.18
<0.001 <0.001 0.056
Shading indicates statistical significance (p < 0.007 with Bonferroni correction). CI ¼ confidence interval.
Fig. 3. This graphic displays the incidence of the occurrence of any adverse event and extended hospital stay when stratified by integer values of preoperative creatinine levels.
Table 4 Impact of preoperative creatinine level on risk for adverse outcomes. Adverse Outcomes
Any Adverse Event Severe Adverse Event Minor Adverse Event Adverse Hospital Metrics Extended Hospital Stay (>3 days) Discharge to Higher Level Care Hospital Readmission
Low Creatinine (Preoperative)
High Creatinine (Preoperative)
Odds Ratio
95% CI
P-value
Odds Ratio
95% CI
P-value
1.44 1.47 1.38
1.25e1.67 1.09e2.00 1.18e1.62
<0.001 0.013 <0.001
1.45 1.27 1.48
1.36e1.54 1.14e1.43 1.38e1.58
<0.001 <0.001 <0.001
1.55 1.31 1.43
1.34e1.79 1.14e1.50 1.08e1.91
<0.001 <0.001 0.014
1.44 1.35 1.40
1.36e1.53 1.29e1.43 1.26e1.55
<0.001 <0.001 <0.001
Shading indicates statistical significance (p < 0.007 with Bonferroni correction). CI ¼ confidence interval.
Finally, with respect to preoperative creatinine, low values were associated with the occurrence of any adverse event, minor adverse events, extended hospital stay, and discharge to higher level care (OR: 1.31e1.55, p < 0.001). Elevated preoperative creatinine values were associated with the occurrence of all adverse outcomes (OR: 1.27e1.48, p < 0.001) (Table 4). 4. Discussion As healthcare continues to adopt advance evidence based treatment algorithms, the incremental value of tests and
procedures is receiving increased scrutiny. This is especially important given the implementation of the voluntary BPCI and mandatory CJR initiatives for joint replacements, which demand healthcare facilities to consider both the quality and cost of treatment [1]. It is estimated that up to 18 billion dollars is spent each year on preoperative laboratory testing and diagnostic studies in the United States [18], and the use of guidelines to inform preoperative evaluation is thought to be key to economical pre-operative preparations [18,19]. Yet, to our knowledge, the clinical value of routine pre-operative laboratory abnormalities remain unclear in modern
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
6
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
elective arthroplasty patients. In this context, the purpose of the present study was to identify trends in the incidence of adverse outcomes across the spectrum of preoperative sodium, BUN, and creatinine levels in a large, national THA patient sample and to characterize correlations between these abnormal lab levels and adverse outcomes. Interestingly, excluding elevated preoperative sodium, the lowest incidence of all studied adverse outcomes occurred well within the traditionally defined normal ranges for sodium, BUN, and creatinine (Figs. 1e3). Additionally, when examining the upper and lower limits of normal ranges for all laboratory values, the highest incidence of any adverse event was within 10% of the local minimum value for all adverse outcomes. These findings support expectations that “normal” basic metabolic panel values of representative markers correlate with unremarkable perioperative courses. In the analyses of preoperative sodium levels, multivariate regressions controlling for patient factors and preoperative BUN and creatinine levels, revealed that abnormally low sodium levels were associated with the occurrence of all studied adverse outcomes (Table 2). Hyponatremia may be associated with increased adverse outcomes due to hypo-osmolality being correlated with impaired immune functioning, which can lead to higher propensities for infectious complications and their associated adverse hospital metrics [20]. Additionally, hyponatremia has been associated with encephalopathy [21], and peripheral nerve malfunction [22], which could respectively lead to attention deficits and gait unsteadiness. This has been documented to occur in patients with mild chronic hyponatremia and could also lead to a variety of adverse events and hospital metrics [23]. Given the high effect sizes of hyponatremia, patients with this finding may benefit from optimization of their sodium values before their planned procedure. Although many mild abnormalities may be due to common medications (thiazide diuretics, angiotensin-converting enzyme inhibitors, etc), it has been suggested that if there is not an apparent cause, investigation of the abnormality would be useful [24]. Interestingly, the same inclusive multivariate regressions revealed no significant associations between abnormally elevated sodium levels and postoperative adverse outcomes (Table 2). Although hypernatremia has also been previously associated with neurological dysfunction and myocardial impairments its hypothesized that the small number of patients in this sample with hypernatremia, in combination with the small range of elevated sodium levels may lead to hypernatremia not being a risk factor in this study [25]. In the analyses of preoperative BUN and creatinine levels, multivariate regressions revealed that both abnormally low and abnormally high preoperative levels were associated with adverse outcomes (Tables 3 and 4). Elevated levels of both of these tests is classically a marker of impaired renal function, however it can also be elevated in a variety of other conditions ranging from dehydration to gastrointestinal blood loss to congestive heart failure [26]. Reduced BUN and creatinine levels, although rare, can be caused by malnutrition, underlying severe liver disease, or syndrome of inappropriate antidiuretic hormone [27]. Regardless of the direction of abnormality, surgeons should consider further workup of high and low preoperative BUN and creatinine levels, as some of the causes are potentially reversible and steps to correct the irregularity may improve a patient's probability of an uncomplicated recovery. For example, studies have found that prophylactic dialysis in patients undergoing coronary bypass with creatinine values higher than 2.5 mg/dL led to decreased perioperative morbidity and mortality in comparison to those not receiving it [28].
Strengths of the present study include the use of the NSQIP dataset. As the patient information is abstracted by trained surgical clinical reviewers using chart review, it is thought to be of higher quality than administratively coded resources, such as the National Inpatient Sample [9]. Additionally, NSQIP provides patient followup for 30 days after surgery, regardless of admission status, allowing for the study of post-discharge adverse events. This is especially important given that nearly all proposed bundled payments require responsibility for care beyond the inpatient hospital stay [1]. Finally, the use of multiple imputation in the present study is novel and important in addressing known concerns for missing laboratory values in NSQIP. Previous literature has suggested that multiple imputation should be the standard for treating missing values [16,17]. Limitations of the present study include that it is impossible to assess actions taken by surgeons in response to abnormal preoperative laboratory values (for example replacing electrolytes in patients with hyponatremia, ordering dialysis for patients with elevated BUN or creatinine, and/or postponing/ canceling surgery) as this would not be captured by NSQIP. Additionally, there is no way to study the associations of other values that are often included in a basic metabolic panel (potassium, chloride, and bicarbonate) on adverse outcomes as they are not reported in the dataset. Despite these limitations, the present study offers the most convincing evidence to date of the utility of preoperative basic metabolic panel testing for the elective THA population. In conclusion, the current study demonstrates that traditionally defined normal ranges of preoperative sodium, BUN, and creatinine generally correlate with a lower incidence of perioperative general health adverse outcomes in THA patients. Further, by defining clear increased odds of adverse events and adverse hospital metrics in the setting of abnormal laboratory studies, even when controlling for patient factors, the current study supports the importance of evaluating these markers in the preoperative THA population. More research will be needed to assess the corrective impact of addressing abnormalities in these laboratories preoperatively. Ethical review committee statement This study has been given an exemption from the senior author's Institutional Review Board under federal regulation 45 CFR 46.101(b) (4). Declaration of Competing Interest The authors have no ownership, patents, or participations with entities whose products or category of products are mentioned in the current manuscript. No funding was received for this study. The authors of the manuscript do declare the following financial conflicts unrelated to the current manuscript: One of the authors (JNG) reports the following financial activities outside the submitted work: past consultancy with Medtronic (Minneapolis, MN, USA), Bioventus (Durham, NC, USA), Stryker (Mahwah, NJ, USA), and TIDI products (Neenah, Wisconsin, USA); and participation in a clinical trial as a sub-investigator for Pfizer (New York City, NY, USA), Spinal Kinetics (Sunnyvale, CA, USA), and Orthofix (Lewisville, TA, USA). All other authors certify that he or she has no commercial associations that might pose a conflict of interest in connection with the submitted article. References [1] Siddiqi A, White PB, Mistry JB, Gwam CU, Nace J, Mont MA, Delanois RE. Effect of bundled payments and health care reform as alternative payment models
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019
N.T. Ondeck et al. / Journal of Orthopaedic Science xxx (xxxx) xxx
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10] [11]
[12]
in total joint arthroplasty: a clinical review. J Arthroplast 2017 Aug;32(8): 2590e7. Froimson MI, Rana A, White Jr RE, Marshall A, Schutzer SF, Healy WL, Naas P, Daubert G, Iorio R, Parsley B. Bundled payments for care improvement initiative: the next evolution of payment formulations: AAHKS Bundled Payment Task Force. J Arthroplast 2013 Sep;28(Suppl. 8):157e65. Alfonso A, Hutzler L, Robb B, Beste C, Blom A, Bosco J. Similar cost savings of bundled payment initiatives applied to lower extremity total joint arthroplasty can Be achieved applying both models 2 and 3. HSS J 2017 Oct;13(3): 267e70. Iorio R, Clair AJ, Inneh IA, Slover JD, Bosco JA, Zuckerman JD. Early results of medicare's bundled payment initiative for a 90-day total joint arthroplasty episode of care. J Arthroplast 2016 Feb;31(2):343e50. Lakomkin N, Sathiyakumar V, Dodd AC, Jahangir AA, Whiting PS, Obremskey WT, Sethi MK. Pre-operative labs: wasted dollars or predictors of post-operative cardiac and septic events in orthopaedic trauma patients? Injury 2016 Jun;47(6):1217e21. Borges A, Torres J, Sao Simao R, Cabral AT, Pinto R. Impact of preoperative analytical values on post-operative mortality rate of intertrochanteric fractures. Acta Med Port 2014 Mar-Apr;27(2):218e22. Bernstein DN, Keswani A, Ring D. Perioperative risk adjustment for total shoulder arthroplasty: are simple clinically driven models sufficient? Clin Orthop Relat Res 2017 Dec;475(12):2867e74. Samuel AM, Lukasiewicz AM, Webb ML, Bohl DD, Basques BA, Davis KA, Grauer JN. ICD-9 diagnosis codes have poor sensitivity for identification of preexisting comorbidities in traumatic fracture patients: a study of the National Trauma Data Bank. J Trauma acute care surg 2015 Oct;79(4): 622e30. McLynn RP, Geddes BJ, Cui JJ, Ondeck NT, Bovonratwet P, Shultz BN, Grauer JN. Inaccuracies in ICD coding for obesity would be expected to bias administrative database spine studies toward overestimating the impact of obesity on perioperative adverse outcomes. Spine 2018 Aug 01;(7):526e32. Program ACoSNSQI. User guide for the 2015 ACS NSQIP participant use data file (PUF). Chicago, IL. 2016. Bovonratwet P, Ondeck NT, Tyagi V, Nelson SJ, Rubin LE, Grauer JN. Outpatient and inpatient unicompartmental knee arthroplasty procedures have similar short-term complication profiles. J Arthroplast 2017 May 18;32(10). Bohl DD, Samuel AM, Basques BA, Della Valle CJ, Levine BR, Grauer JN. How much do adverse event rates differ between primary and revision total joint arthroplasty? J Arthroplast 2016 Mar;31(3):596e602.
7
[13] Bohl DD, Fu MC, Golinvaux NS, Basques BA, Gruskay JA, Grauer JN. The "July effect" in primary total hip and knee arthroplasty: analysis of 21,434 cases from the ACS-NSQIP database. J Arthroplast 2014 Jul;29(7):1332e8. [14] Menard SW. Applied logistic regression analysis. Sage Publications; 1995. [15] Hair JF. Multivariate data analysis. Prentice Hall; 1998. [16] Parsons HM, Henderson WG, Ziegenfuss JY, Davern M, Al-Refaie WB. Missing data and interpretation of cancer surgery outcomes at the American College of surgeons national surgical quality improvement program. J Am Coll Surg 2011 Sep;213(3):379e91. [17] Ondeck NT, Fu MC, Skrip LA, McLynn RP, Su EP, Grauer JN. Treatments of missing values in large national data affect conclusions: the impact of multiple imputation on arthroplasty research. J Arthroplast 2018 Mar;33(3):661e7. [18] Richman DC. Ambulatory surgery: how much testing do we need? Anesthesiol Clin 2010 Jun;28(2):185e97. [19] Fischer SP. Cost-effective preoperative evaluation and testing. Chest 1999 May;115(Suppl. 5):96se100s. [20] Jensen AG, Wachmann CH, Poulsen KB, Espersen F, Scheibel J, Skinhoj P, Frimodt-Moller N. Risk factors for hospital-acquired Staphylococcus aureus bacteremia. Arch Intern Med 1999 Jul 12;159(13):1437e44. [21] Hoorn EJ, Zietse R. Hyponatremia and mortality: moving beyond associations. Am J Kidney Dis 2013 Jul;62(1):139e49. [22] Aranyi Z, Kovacs T, Szirmai I, Vastagh I. Reversible nerve conduction slowing in hyponatremia. J Neurol 2004 Dec;251(12):1532e3. [23] Renneboog B, Musch W, Vandemergel X, Manto MU, Decaux G. Mild chronic hyponatremia is associated with falls, unsteadiness, and attention deficits. Am J Med 2006 Jan;119(1):71 e1e8. [24] Klinck J, McNeill L, Di Angelantonio E, Menon DK. Predictors and outcome impact of perioperative serum sodium changes in a high-risk population. Br J Anaesth 2015 Apr;114(4):615e22. [25] Adrogue HJ, Madias NE. Hypernatremia. N Engl J Med 2000 May 18;342(20): 1493e9. [26] Jafari SM, Huang R, Joshi A, Parvizi J, Hozack WJ. Renal impairment following total joint arthroplasty: who is at risk? J Arthroplast 2010 Sep;25(Suppl. 6): 49e53. e1-2. [27] Longo D, Fauci A, Kasper D, Hauser S, Jameson J, Loscalzo J. Harrison's principles of internal medicine. 18th ed. McGraw-Hill Education; 2011. [28] Durmaz I, Yagdi T, Calkavur T, Mahmudov R, Apaydin AZ, Posacioglu H, Atay Y, Engin C. Prophylactic dialysis in patients with renal dysfunction undergoing on-pump coronary artery bypass surgery. Ann Thorac Surg 2003 Mar;75(3): 859e64.
Please cite this article as: Ondeck NT et al., Preoperative laboratory testing for total hip arthroplasty: Unnecessary tests or a helpful prognosticator, Journal of Orthopaedic Science, https://doi.org/10.1016/j.jos.2019.09.019